Streamline Your AI Agent Development with the Nova Act IDE Extension
Introduction to the Nova Act IDE Extension
The development of AI agents has traditionally involved a fragmented workflow, requiring developers to constantly switch between their Integrated Development Environment (IDE) for coding and a browser for testing and debugging. This constant context-switching not only disrupts the creative flow but also significantly slows down the development process. Amazon Web Services (AWS) addresses this challenge with the introduction of the Nova Act IDE extension. This powerful tool integrates the entire AI agent development lifecycle—from initial ideation and script generation to granular customization and robust testing—directly within your familiar IDE.
Key Features for Accelerated Development
Natural Language Script Generation
One of the most transformative features of the Nova Act extension is its ability to generate initial agent scripts from natural language descriptions. Instead of starting with boilerplate code, you can simply describe your desired automation task in plain English. For example, you can articulate a need like, "I need an agent that logs into a customer portal, searches for unresolved tickets, and updates their status based on completion criteria." The extension then generates a functional script that handles authentication, navigation, search, and status updates, providing a solid foundation upon which to build.
Notebook-Style Builder Mode
The Nova Act extension introduces a notebook-style Builder Mode that breaks down complex automation scripts into modular, manageable cells. This approach allows for incremental development and precise testing. You can execute your script cell by cell, making small changes and testing them individually before proceeding. This modularity is invaluable for complex workflows, enabling developers to test different approaches, tweak parameters, and validate results in real-time without re-running the entire script. You can add cells, reorder them, and interleave act() statements with regular Python code to construct robust agents incrementally. Within this mode, you can call nova.act() multiple times within a single session, with each session corresponding to a browser instance. To start a fresh browser session, you can either call nova.stop() followed by nova.start(), or utilize the Restart Notebook button.
Live Debugging and Action Viewer
Debugging is significantly enhanced with the Nova Act extension's live debugging capabilities. It unifies your code editor, browser view, and execution logs into a single, cohesive experience. You can observe your agent executing in real-time, monitoring its decision-making process, and understanding exactly what it sees and why it makes certain choices. This allows for pausing execution at any point to make adjustments and then resuming without starting over. Furthermore, the Action Viewer provides a detailed analysis of completed runs. You can examine individual act() statements or the entire end-to-end workflow. The ability to compare multiple runs side-by-side is particularly powerful for identifying improvements or regressions without manually sifting through separate log files.
Seamless Chat and Builder Handoff
The extension offers a seamless integration between chat-based script generation and the Builder Mode. You can initiate script creation through a chat interface, where three workflow modes are available: Ask (to generate scripts from natural language), Edit (to refine existing scripts), and Agent (to run, monitor, and interact with the agent). Once a script is generated or refined in chat, you can convert it into a Python file in Builder Mode with a single click, facilitating a smooth transition between natural language prototyping and detailed scripting.
Contextual Information and Templates
To further aid the agent's understanding and execution, you can provide Context. This includes relevant information about active documents, instructions, problems, or additional Model Context Protocol (MCP) resources. You can also include a screenshot of the current window, which helps the agent grasp specific requirements for the automation task. Additionally, the Nova Act extension provides a set of predefined templates accessible by typing "/" in the chat. These templates offer quick-start solutions for common web tasks, such as /shopping for e-commerce automation, /extract for data extraction, /search for information gathering, /qa for quality assurance, and /formfilling for data entry.
Getting Started with Nova Act
Installation
To begin, install the Nova Act extension from your IDE's extension manager. For Visual Studio Code, navigate to Extensions, search for "Nova Act," and select Install. The extension is also available for Cursor and Kiro, with plans for additional IDE support.
API Key Configuration
After installation, you will need to configure your Nova Act API key. This can be done by opening the Command Palette (Cmd+Shift+P or Ctrl+Shift+P) and selecting Set API Key. Follow the instructions to obtain and enter your API key.
Exploring Builder Mode and Chat
Once your API key is set, you can immediately start using Builder Mode. This notebook-style interface allows you to break down your automation into discrete cells, test each step, and debug effectively. Alternatively, you can leverage the chat interface to generate scripts using natural language. After generating a script via chat, you can seamlessly transition to Builder Mode to refine and test it further.
Technical Requirements and Considerations
Supported IDEs and Python Version
At launch, the Nova Act extension supports Visual Studio Code, Cursor, and Kiro. It requires Python version 3.10 or higher. Ensure your operating system is macOS Sierra+, Ubuntu 22.04+, or Windows 10+.
Open Source and Pricing
The Nova Act extension is an open-source project available under the Apache 2.0 license, encouraging community contributions and customization. Importantly, the extension is available at no charge.
Troubleshooting Port Usage
If you encounter a StartFailed error, it may be due to a port conflict. The Nova Act extension uses local ports 9222 (for Chrome DevTools) and 8001 (for internal communication). If these ports are in use, the extension might not start correctly. A quick fix is to use the Restart Notebook button in Builder Mode. If that doesn't resolve the issue, you may need to manually identify and terminate processes using these ports on macOS/Linux (using lsof and kill) or Windows (using netstat and taskkill).
Conclusion
The Nova Act IDE extension represents a significant leap forward in AI agent development. By integrating natural language processing, modular scripting, and live debugging into a single, cohesive IDE experience, it empowers developers to build, customize, and validate production-grade agent scripts faster and more efficiently than ever before. Whether you are prototyping with natural language, refining with modular scripting, or validating with local testing, Nova Act provides a full-stack solution without leaving your IDE.
AI Summary
The Nova Act IDE extension, built upon the Amazon Nova Act SDK, significantly accelerates AI agent development by consolidating the entire lifecycle—from ideation to production—within a unified IDE interface. This eliminates the need to switch between multiple tools, reducing context-switching and enhancing developer velocity. Key features include natural language-based script generation, a notebook-style builder mode for modular development and step-by-step testing, and live debugging that provides real-time insights into the agent's decision-making process. The extension supports various IDEs like Visual Studio Code, Cursor, and Kiro, and offers predefined templates for common tasks such as shopping, data extraction, and form filling. Its open-source nature and no-cost availability further encourage community contributions and widespread adoption, positioning it as a comprehensive IDE for AI agent development.